基于深度学习算法的机器自动翻译质量评估模型

Quality evaluation model of MT based on deep learning algorithm

以实现机器自动翻译质量准确评估为研究目标,构建了基于深度学习算法的机器自动翻译质量评估模型。通过基于深度学习的机器自动翻译语言信息提取方法,在无监督学习阶段、有监督学习阶段中,利用降噪自动编码机对双语词实施无监督学习,重构并获取机器自动翻译样例中的语言向量特征,将机器自动翻译信息导入到双语词中,优化语言向量特征提取效果;将机器自动翻译语言向量特征导入基于深度学习的译文质量评估模型中,实现机器自动翻译质量评估。实验结果表明,所构建模型可准确评估机器自动翻译质量,且翻译句型、语句数量对该模型评估性能不存在负面干扰,模型使用效果较好。

In order to realize the accurate evaluation of MT quality,a model of MT quality evaluation based on deep learning algorithm is constructed.In the unsupervised learning stage and supervised learning stage,the noise reduction automatic coder is used to implement unsupervised learning of bilingual words,reconstruct and obtain the language vector features in the sample of automatic machine translation,import the automatic machine translation information into the bilingual words,and optimize the extraction effect of language vector features.The vector features of dynamic translation language are introduced into the translation quality evaluation model based on deep learning to realize automatic translation quality evaluation.The experimental results show that the model can accurately evaluate the quality of automatic machine translation,and there is no negative interference on the evaluation performance of the model due to the sentence pattern and the number of sentences.